[mesa-users] Problems with 'newton iters' determined timesteps

[Wonseok Chun]

I am currently running He Zams evolution with semiconvection. However, as
calculation progress on, I have issues with small timestep, about 0.01yr.

First issue appears near the start of calculation, and timestep is reduced
from 'newton iters'.
I was able to raise timestep by rasing newton_iteration_limit.

However, timestep is again reduced at age around log_age~5.29. Timestep is
again forcibly reduced from failure of convergence from newton_hydro check,
and I presume this is due to steep gradiant of c12 reaching surface, while
there is still large enough mass loss from wind.
(Mass loss reduces convective core size, which in turn makes c12 abundance
gradient.)

Similar problems occurs in 10M model too, but without apparent reason.

Initial model used is 20M, he4=0.98, Z=0.02 with dxdt_nuc_factor=0
relaxtion.
ZAMS model was constructed by following steps.

- make pre_ms with y=0,z=0, 20m
- relax y to y=1
- calculate until central he4=0.98, with dxdt_nuc_factor=0.01
- set uniform initial abundance - h1,h2,he3=0, he4=0.98

And additional parameters are

- Wind : dutch
- semiconvection : Ledoux, a=0.01
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.mesastar.org/pipermail/mesa-users/attachments/20140110/a9c0deae/attachment.html>